Specify Embedding dimension for multimodal input

This code sample shows how to specify a lower embedding dimension for text and image inputs.

Explore further

For detailed documentation that includes this code sample, see the following:

• Get multimodal embeddings

Code sample

Go

Before trying this sample, follow the Go setup instructions in the Vertex AI quickstart using client libraries. For more information, see the Vertex AI Go API reference documentation.

To authenticate to Vertex AI, set up Application Default Credentials. For more information, see Set up authentication for a local development environment.

import (
	"context"
	"encoding/json"
	"fmt"
	"io"

	aiplatform "cloud.google.com/go/aiplatform/apiv1beta1"
	aiplatformpb "cloud.google.com/go/aiplatform/apiv1beta1/aiplatformpb"
	"google.golang.org/api/option"
	"google.golang.org/protobuf/encoding/protojson"
	"google.golang.org/protobuf/types/known/structpb"
)

// generateWithLowerDimension shows how to generate lower-dimensional embeddings for text and image inputs.
func generateWithLowerDimension(w io.Writer, project, location string) error {
	// location = "us-central1"
	ctx := context.Background()
	apiEndpoint := fmt.Sprintf("%s-aiplatform.googleapis.com:443", location)
	client, err := aiplatform.NewPredictionClient(ctx, option.WithEndpoint(apiEndpoint))
	if err != nil {
		return fmt.Errorf("failed to construct API client: %w", err)
	}
	defer client.Close()

	model := "multimodalembedding@001"
	endpoint := fmt.Sprintf("projects/%s/locations/%s/publishers/google/models/%s", project, location, model)

	// This is the input to the model's prediction call. For schema, see:
	// https://cloud.google.com/vertex-ai/generative-ai/docs/model-reference/multimodal-embeddings-api#request_body
	instance, err := structpb.NewValue(map[string]any{
		"image": map[string]any{
			// Image input can be provided either as a Google Cloud Storage URI or as
			// base64-encoded bytes using the "bytesBase64Encoded" field.
			"gcsUri": "gs://cloud-samples-data/vertex-ai/llm/prompts/landmark1.png",
		},
		"text": "Colosseum",
	})
	if err != nil {
		return fmt.Errorf("failed to construct request payload: %w", err)
	}

	// TODO(developer): Try different dimenions: 128, 256, 512, 1408
	outputDimensionality := 128
	params, err := structpb.NewValue(map[string]any{
		"dimension": outputDimensionality,
	})
	if err != nil {
		return fmt.Errorf("failed to construct request params: %w", err)
	}

	req := &aiplatformpb.PredictRequest{
		Endpoint: endpoint,
		// The model supports only 1 instance per request.
		Instances:  []*structpb.Value{instance},
		Parameters: params,
	}

	resp, err := client.Predict(ctx, req)
	if err != nil {
		return fmt.Errorf("failed to generate embeddings: %w", err)
	}

	instanceEmbeddingsJson, err := protojson.Marshal(resp.GetPredictions()[0])
	if err != nil {
		return fmt.Errorf("failed to convert protobuf value to JSON: %w", err)
	}
	// For response schema, see:
	// https://cloud.google.com/vertex-ai/generative-ai/docs/model-reference/multimodal-embeddings-api#response-body
	var instanceEmbeddings struct {
		ImageEmbeddings []float32 `json:"imageEmbedding"`
		TextEmbeddings  []float32 `json:"textEmbedding"`
	}
	if err := json.Unmarshal(instanceEmbeddingsJson, &instanceEmbeddings); err != nil {
		return fmt.Errorf("failed to unmarshal JSON: %w", err)
	}

	imageEmbedding := instanceEmbeddings.ImageEmbeddings
	textEmbedding := instanceEmbeddings.TextEmbeddings

	fmt.Fprintf(w, "Text embedding (length=%d): %v\n", len(textEmbedding), textEmbedding)
	fmt.Fprintf(w, "Image embedding (length=%d): %v\n", len(imageEmbedding), imageEmbedding)
	// Example response:
	// Text Embedding (length=128): [0.27469793 -0.14625867 0.022280363 ... ]
	// Image Embedding (length=128): [0.06225733 -0.040650766 0.02604402 ... ]

	return nil
}

Python

Before trying this sample, follow the Python setup instructions in the Vertex AI quickstart using client libraries. For more information, see the Vertex AI Python API reference documentation.

To authenticate to Vertex AI, set up Application Default Credentials. For more information, see Set up authentication for a local development environment.

import vertexai

from vertexai.vision_models import Image, MultiModalEmbeddingModel

# TODO(developer): Update & uncomment line below
# PROJECT_ID = "your-project-id"
vertexai.init(project=PROJECT_ID, location="us-central1")

# TODO(developer): Try different dimenions: 128, 256, 512, 1408
embedding_dimension = 128

model = MultiModalEmbeddingModel.from_pretrained("multimodalembedding@001")
image = Image.load_from_file(
    "gs://cloud-samples-data/vertex-ai/llm/prompts/landmark1.png"
)

embeddings = model.get_embeddings(
    image=image,
    contextual_text="Colosseum",
    dimension=embedding_dimension,
)

print(f"Image Embedding: {embeddings.image_embedding}")
print(f"Text Embedding: {embeddings.text_embedding}")

# Example response:
# Image Embedding: [0.0622573346, -0.0406507477, 0.0260440577, ...]
# Text Embedding: [0.27469793, -0.146258667, 0.0222803634, ...]

What's next

To search and filter code samples for other Google Cloud products, see the Google Cloud sample browser.